Legal claims defining the scope of protection, as filed with the USPTO.
2. The method of claim 1, wherein the extracting of the reference features for the target object, the object candidate, and the tracking area comprises extracting the reference features on the basis of a Siamese neural network structure including a plurality of identical neural networks.
3. The method of claim 2, wherein the plurality of identical neural networks include a convolutional neural network (CNN) performing a convolutional operation.
4. The method of claim 3, wherein the CNN includes a fully convolutional network structure.
6. The method of claim 1, wherein the allocating of the object candidate to the target object on the basis of the evaluated similarity comprises allocating, to the target object, an object candidate corresponding to an area exhibiting the highest similarity to the target object in a similarity map.
7. The method of claim 1, wherein the tracking of the target object on the basis of the location of the allocated object candidate comprises tracking the target object by applying a phase value of an object corresponding to the location of the allocated object candidate to the target object.
8. The method of claim 1, further comprising training the integrated similarity neural network by inputting a location of the target object, which is changed in consideration of a result of the tracking, to the integrated similarity neural network.
10. The multi-object tracking apparatus of claim 9, wherein the command to extract the reference features for the target object, the object candidate, and the tracking area comprises a command to extract the reference features on the basis of a Siamese neural network structure including a plurality of identical neural networks.
11. The multi-object tracking apparatus of claim 10, wherein the plurality of identical neural networks include a convolutional neural network (CNN) performing a convolutional operation.
12. The multi-object tracking apparatus of claim 11, wherein the CNN includes a fully convolutional network structure.
14. The multi-object tracking apparatus of claim 9, wherein the command to allocate the object candidate to the target object on the basis of the evaluated similarity comprises a command to allocate, to the target object, an object candidate corresponding to an area exhibiting the highest similarity to the target object in a similarity map.
15. The multi-object tracking apparatus of claim 9, wherein the command to track the target object on the basis of the location of the allocated object candidate comprises a command to track the target object by applying a phase value of an object corresponding to the location of the allocated object candidate to the target object.
16. The multi-object tracking apparatus of claim 9, wherein the command further comprises a command to train the integrated similarity neural network by inputting a location of the target object, which is changed in consideration of a result of the tracking, to the integrated similarity neural network.
Unknown
July 4, 2023
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.